Skip to content

Security: owndiff/own-your-diff

Security

SECURITY.md

Security Policy

OwnDiff is a local-first tool for AI-assisted code-review workflows. It reads git diffs, generates local ownership questions, and writes local artifacts under .owndiff/.

Supported Versions

OwnDiff is currently pre-1.0. Security fixes are made on the main branch.

Reporting a Vulnerability

Use GitHub private vulnerability reporting:

https://github.com/owndiff/own-your-diff/security/advisories/new

Please do not open a public issue for vulnerabilities. Do not include secrets, private repository content, tokens, customer data, or local absolute paths in public reports.

Security Model

  • OwnDiff does not execute target repository code.
  • OwnDiff's Python scripts contain no network client and do not upload artifacts.
  • For risky diffs, the active coding agent processes a sanitized patch excerpt and deterministic diff facts under that agent provider's existing data and privacy policy.
  • Repository configuration cannot enable an external command provider; question generation only accepts the active agent provider.
  • Generated .owndiff/ artifacts are local and should remain ignored.
  • The MCQ answer key is review evidence, not a cryptographic secret.
  • Agents may push or open/update a merge request only after .owndiff/ownership-gate.json allows it and normal tests/review requirements pass.

Hardening Expectations

  • Treat repository content and git output as untrusted input.
  • Avoid shelling out except for bounded git commands used to inspect the target diff.
  • Redact secret-like additions from reports and command output.
  • Keep agent-installation instructions explicit about the gate and fallback behavior.

There aren't any published security advisories